Gendered Outcome Bias

How to cite

Giffin, Erin; Landsman, Rachel; Halladay, Brianna. 2024. "Gendered Outcome Bias." AEA RCT Registry. AEARCTR-0014905. https://doi.org/10.1257/rct.14905-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Erin Giffin* , Rachel Landsman , Brianna Halladay

* Primary investigator

Registry ID
AEARCTR-0014905
DOI
https://doi.org/10.1257/rct.14905-1.0
First registered
Last updated
Status
In development
Trial dates
2024-12-04 – 2025-05-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14905

Abstract

Recent empirical studies have shown differences in the punishment and rewarding of men and women following uncertain outcomes that are due to both luck and the agent’s actions. Many studies have shown that men tend to be rewarded following unexpected good outcomes while women tend to be punished more frequently following unexpected bad outcomes. It is unclear whether these gender differences are due to a difference in how male versus female agents' outcomes are attributed to their actions (outcome bias) or whether they simply stem from a difference in a desire to punish/reward male versus female agents. We build on this literature by designing an experimental study to disentangle these competing mechanisms. We ask whether gender differences in rewards/punishments are due to differences in outcome bias depending on agent gender or differences in punishment preferences dependent on the agent’s gender.

Experimental design

This is a two-sided experiment where participants are assigned to the role of either "principal" or "agent". Agents perform a task, and the outcome of the task is determined by their choice as well as chance ("luck"). Principals receive a bonus based on the outcome of the task. Principals observe the agent's gender as well as some information related to the task (our treatment conditions systematically vary which information is revealed). After seeing this information, principals can choose to punish the agent by reducing the agent's payment at a cost to themselves.

Design details

Randomization method
Randomization will be done within the Qualtrics survey, using the Qualtrics randomizer tool in survey flow. After completing the consent form, participants will be randomly assigned to one of the treatments and directed to the version of the survey that aligns with that condition.
Randomization unit
Individual
Clusters
The design is not clustered, so the number of observations is equal to the number of clusters (1200 participants who are assigned to the role of "principal").
Observations
1200 participants
Arms
800 participants in the "outcome and choice revealed" condition (400 participants in "agent" role and 400 participants in "principal" role), 800 participants in the "outcome hidden" condition (400 participants in "agent" role and 400 participants in "principal" role), 800 participants in the "choice hidden" condition (400 participants in "agent" role and 400 participants in "principal" role)
Primary outcomes
Principals' punishment decision

Cited by

No citation data available for this registration yet.

Keywords